Psychometric deviance in offspring at risk for schizophrenia: II. Resolving heterogeneity through admixture analysis.

The longitudinal and prospective study of offspring at risk for schizophrenia is complicated by within-group heterogeneity in liability, as only a subgroup of those at risk will ultimately become affected. Here, we attempt to resolve such heterogeneity in the New York High-Risk Project by conducting an admixture analysis of values on a psychometric index of liability to schizophrenia derived from the Minnesota Multiphasic Personality Inventory (MMPI). We fit mixtures of components to the overall distribution in 171 children from three criterion groups: offspring at risk (HR) for schizophrenia, psychiatric comparison (PC) offspring at risk for affective illness, and normal comparison (NC) offspring not at increased risk for psychiatric morbidity. The distribution of psychometric scores was bimodal, and separation of two latent classes showed that there is a valid and nonarbitrary distinction between a subgroup of MMPI-deviant (primarily HR) offspring and a larger homogeneous group of MMPI-nondeviant HR, PC, and NC subjects. While continued followup is required to demonstrate a correspondence between these two classes and an underlying taxonomy of liability to schizophrenia, our findings demonstrate the utility of objective psychometric measurement and admixture analysis for resolving within-group heterogeneity in high-risk research. The wider implications of including our MMPI indicators in other genetic investigations of schizophrenia are discussed.